Torn Between 2012 MP or nMP for Audio Recording

[cinealta]

However perhaps you should compare the base nMP ($3k) with a hex core BTO upgrade ($?, maybe $200-400?). With the D500 hex core you are paying for GPU you don't need AFAIK, same with the 7950.

The nMP 4-core ($3K) with 6-core BTO upgrade is an interesting option. Deconstruct60 estimates that to be an additional +$500 (earlier in this thread). It may still be an approx $1,300 difference between the cMP & nMP (with $600 bare Thunderbolt enclosure). Comparing apples-to-apples, I would then not price the 7950 GPU into the cMP making it approx $2,700.

The 4-core nMP, without 6-core BTO upgrade, is actually slower than the 2012 6-core cMP. Also, AFAIK Logic Pro X can take advantage of 8 physical cores.

 

[Cubemmal]

The nMP 4-core ($3K) with 6-core BTO upgrade is an interesting option. Deconstruct60 estimates that to be an additional +$500 (earlier in this thread). It may still be an approx $1,300 difference between the cMP & nMP (with $600 bare Thunderbolt enclosure). Comparing apples-to-apples, I would then not price the 7950 GPU into the cMP making it approx $2,700.

The 4-core nMP, without 6-core BTO upgrade, is actually slower than the 2012 6-core cMP. Also, AFAIK Logic Pro X can take advantage of 8 physical cores.

Sounds good. I also would consider the noise, heat and size differences. I am a musician and have done some audio engineering, those aspects are worth their weight IMO. My 2009 is quiet, but I'm sure the nMP will be quieter.

 

[deconstruct60]

U
I'm concerned that the USB 3.0 protocol may not be optimum for clocked digital audio streaming, ie isochronous transfer. AFAIK, USB 3.0 throughput can vary with the quality of the host/device controller, encode/decode error correction, varying file size, non-direct memory access transfer etc.

USB has isochronous capability. If isolated on the SuperSpeed bus ( which is separate from the legacy cabable USB 2.0 bus ) there is enough bandwidth to slice up for reasonable large workloads.

The Presonus site has a webpage with summary of "the Pros and Cons of FireWire vs USB." Although it's for USB2, the basics remain the same.

The highlights are:

* FireWire streams data rather than packets data (as USB does).

That seems far more an attempt as a laymay simplification about USB 2.0 synchronization and polling than about stream and packets. Firewire is network which data needs to be directed and there are multiple sources and destinations. Can't locked down all the wires to just two points. All the more so when may have to pass through multiple points to get between the source and destination. If timeslicing wire utilization there is going to be packets.

There are packets in Firewire isochronus:

http://msdn.microsoft.com/en-us/library/windows/hardware/ff537380(v=vs.85).aspx

What USB 2.0 has that the SuperSpeed bus of USB 3.0 doesn't have is the CPU polling the devices for data. So one of the "can't transmit" constraints is can't "speak unless spoken to". That very signficiantly changed with USB 3.0. SuperSpeed. It isn't the same and the old FW vs. USB 2.0 FAQs about that are more so dogma at this point than reality.

* FireWire is typically dedicated for audio/video purposes, and isn't in use by other services on your computer.

There is about zero reason why this can't be done on USB. Especially if using Thunderbolt to add yet another USB controller to the system. This isn't a property of the networks. It is property of isolating the networks. If get an independent USB 3.0 controller and only put USB 3.0 SuperSpeed devices on it then won't be USB 2.0 interrupts and constraints on that specific network.

The USB controller you get for "free" with the chipset coupled to the CPU. Yeah, that one has multiple uses and can't be well isolated from various network loadings. But that doesn't have to be the only one.